The process was a HR screen to validate experience, background, and salary information, followed by a Tech Challenge with coding and questions. There was a System Design round for designing a streaming service and questions, and a Stakeholder round focusing on cultural proposing.

1. What is the disadvantage of using very short segments (chunks) in HLS? Answer: Increase in HTTP request overhead. Although it reduces latency, the player needs to request new files much more frequently, which generates an excessive load of metadata processing on the network and the server.

2. Why use UDP (via QUIC or WebRTC) instead of TCP for real-time? Answer: To avoid Head-of-Line Blocking. TCP guarantees the delivery of every packet and waits for retransmissions if something fails, which causes "freezing". In real-time streaming, it is preferable to lose a small frame than to delay the entire transmission waiting for old data.

3. What is the role of Apache Kafka in an ingestion pipeline? Answer: Decoupling and Resilience. Kafka acts as a buffer that receives data streams and allows transcoding and analysis services to process tasks asynchronously, better handling sudden traffic spikes.

4. Difference between "Push" and "Pull" architecture in ingestion? Answer:

• Push: The encoder actively sends (pushes) data to the server (e.g., RTMP).
• Pull: The destination server requests and fetches data from the original source when needed (common in content delivery networks).

Would you like me to develop a study guide?